About the AIPNET project
See the working paper for our main results and discussion.
The AI-generated Production Network (AIPNET) aims to enhance our understanding of international trade and economic structures by mapping the relationships between products in the global economy. Using generative AI technology, we’ve developed a network of over 5,000 products connected by their input-output relationships in the production process.
Our “build-prune” methodology combines multiple calls to LLMs to first identify and then verify product relationships, creating a detailed view of global production chains. This approach can be deployed to a wide range of other contexts, where nodes are known but their connections are unknown.
This network allows us to track changes in how products and countries connect in the 21st century global economy.
Through AIPNET, we’ve examined production network spillovers by studying real-world events like the 2017 Qatar blockade. These findings contribute to research possibilities in onshoring, industrial policy, and the evolution of global trade patterns.